OLAP workloads

Online Analytical Processing (OLAP) workloads are characterized by complex queries that analyze large amounts of historical data to identify patterns, trends, and insights. These workloads often requiring longer execution times than simpler transactional workloads. This is because OLAP queries typically involve aggregating, summarizing, and exploring vast datasets rather than modifying individual records.

DBtune can effectively optimize these complex OLAP workloads, but it requires adjustments to capture their unique characteristics. Since OLAP queries can run for extended periods, DBtune needs sufficient time to observe each query's full execution. This ensures accurate performance analysis and effective tuning recommendations.

To accommodate these longer queries, you can customize the duration of each DBtune iteration. By default, DBtune uses a shorter iteration time optimized for transactional workloads. However, for OLAP systems, you'll need to extend this to encompass the longest running queries.

This adjustment can be made within the src/default_config.yml DBtune configuration file. Locate the measurement_time parameter, currently set to 180 seconds (3 minutes). Increase this value to a multiple of your longest query's execution time. For instance, if your longest query takes 1000 seconds (approximately 17 minutes), consider setting measurement_time to 5000 seconds (approximately 83 minutes) to provide ample observation time.